Forecasting

Ternary Forecasting estimates future cloud spend from historical cost patterns using daily, weekly or monthly granularity. Available lookback and lookahead choices depend on the selected granularity and available history.

Forecasting is commonly used to anticipate budget risk, understand cost growth trends, and support financial planning decisions.

Ternary Forecasting can be accessed from the Cost Assist section in the left-hand navigation menu.

If your navigation includes Plan, open Plan → Forecasting. Forecasting requires Cost Assist access and a configured integration for the organization.

How does cloud cost forecasting work in Ternary

Forecasting in Ternary analyzes historical spend patterns and projects future costs based on the selected time range, granularity, and forecasting parameters. Here’s how you can customize it:

Step 1: Select the forecast scope

Forecasting includes cost data from configured cloud service providers by default and groups it by provider. Use the grouping picker to select another available breakdown, such as category or service description.

Step 2: Choose groupings, measure, and granularity

Groupings and measures can be adjusted to refine analysis by service, category, or other supported dimensions. Forecasts can be viewed at daily, weekly, or monthly granularity.

Step 3: Configure time ranges

Choose a lookback period from the available history options. Monthly granularity excludes the 30-day and 60-day lookbacks. Fixed lookbacks longer than the measured available history are hidden; Max Range appears only when applicable.

Choose a lookahead for the selected granularity:

Step 4: Interpret forecast visualizations

Actual spend for the current period is displayed in a darker shade, while forecasted spend appears in a lighter shade for visual separation.

Users can switch between two views inside the graph:

• Forecast with Lookback (default)

• Forecast Only views

Limited history can return actual spend without a future prediction, or predictions for only some groups. Forecasting requires at least three prepared historical observations for a series. It uses complete periods, excluding the leading partial period and the current partial period, so three calendar periods of visible data do not necessarily provide three usable observations.

When no prediction is returned, the aggregate estimated current-month total can fall back to month-to-date actuals. Groups without predictions can also retain actual spend in their estimates. Do not interpret an actuals-only value as a complete month-end prediction; compare the available historical and forecast portions before using it for planning.

Step 5: Review KPIs and detailed data

Above the chart, key performance indicators (KPIs) provide quick comparisons, including current month-to-date spend vs. previous month-to-date spend, estimated total spend vs. previous total spend, and forecasted spend vs. the average of the previous 3 months.

Summary cards and table values remain based on the current month, regardless of chart granularity or lookahead. Estimated Delta Spend is the estimated current-month total minus the previous month’s total; a 12-month chart does not make these 12-month totals.

Step 6: Export

Select Export after table data has loaded to download the forecasting CSV. The export contains the full dataset for the selected forecast scope, including its provider, applied forecast filters, measure and groupings.

Table search, filters created by clicking a dimension value, sorting and pagination only affect the displayed table. They do not narrow or reorder the CSV. If you need an export for a narrower forecast scope, change and apply the forecast filters before exporting.

The spend columns in the CSV have the same current-month basis as the table.